= 0,1 
ts 
Lehrte, 
7: Area 2| 
Fig. 
     
   
  
      
    
  
  
  
   
    
     
   
     
   
  
  
  
  
  
    
HIGH ACCURATE GEOMETRIC RECTIFICATION 
- NECESSITY AND REALIZATION - 
Dipl.-Ing. Rainer Hóssler 
Lehrstuhl für Photogrammetrie 
Technische Universitat München 
Arcisstr. 21, D-8000 München 2 
INTRODUCTION AND STATEMENT OF THE PROBLEM 
In the early stage interpretation of remote sensing imagery 
was restricted to the use of single aerial photographs or 
to a synoptic consideration of multispectral or multi- 
temporal remote sensing imagery. Out of understandable 
reasons the users were not willing to content themselves 
with those simple interpretation methods. To make better 
use of the full information high developed data collecting 
systems provide, it was necessary to improve interpretation 
methods. This became possible by using analogous and 
digital image processing systems. Those developments not 
only managed to work up more information but also allowed 
to obtain better results. 
In the beginning of image processing geometry wasn't a very 
important object, but with the development of modern 
methods high geometric fidelity became a precondition for 
their useful application. In particular change detection is 
such a new and high developed interpretation method, which 
requires high geometric accuracy to guarantee good results. 
Satisfying results however can be received only, if the 
detected changes are not caused by wrong or less accurate 
geometry. 
The methods for geometric rectification of remote sensing 
data existing up to now can be divided into two groups. The 
more simple methods fit in the data into a given pattern of 
control points according to a certain interpolation model. 
Possible models applied up to now are similarity transfor- 
mations, affin transformations, linear prediction with